See subject.
Thanks.
Out of curiosity, what are the experiments that involve firing a single photon?
This is hard to do, since most sources do not produce single photons on command. Single photon experiments are usually done with a source that has been attenuated to the point where the probability is small that more than one photon will interact with the measurement at the same time. This inevitably means that the time of arrival of the photons will be, to some extent, unpredictable.
There is such a thing as a single photon source, at least theoretically. People are trying to make them, but the technology hasn’t been perfected, since highly correlated quantum states are fragile in the macroscopic world.
IIRC from physics class, that would be the double slit experiment.
The double slit experiment can be performed with a low rate photon source (usually emitting about 10[sup]5[/sup] photons/second) such that there is only one photon on the path at a given time. When doing so, you can see that even single photons demonstrate self-interference stochastically although the photon is obviously detected at a discrete locus.
Single photon emission is possible by using unamplified stimulated emission of an excited medium (somewhat like a laser but without the reflective “chamber”) but it requires very fine control and the photon then has a very random momentum vector (e.g. it goes off in a random direction). Directional single photon emission using solid state devices is very valuable for quantum computation and signals encryption, but although this has been demonstrated in experiement there are, as far as I know, no production-ready devices that can perform single photon emission.
Stranger
It was precisely the double slit experiment I was thinking of. But I’m not understanding “low rate photon source,” which in my mind is begging the question. Sources sorted by brute measurement? Application-prediction via Einstein photo-electric effect?
Now, when someone (Feynman, in a to-laymen lecture) says, “say 10 or photons reflect here (or he says ”15 photons +_5”), this is a figure of speech, right?
Or is the second verbalization more accurate, ultimately?
Can you give me a cite or two that I might understand regarding these apparatuses that I might further explore–ie at the scientific level of these posts? 
The most common source in quantum optics is probably one operating through spontaneous parametric down conversion – basically, you shine a strong laser on a special type of (nonlinear) crystal, commonly beta-barium borate, and at random times, it emits two entangled photons along specific trajectories. One, the signal, is then detected directly, which gives you the knowledge that the other, the idler, is in the apparatus (say, an interferometer of some sort). This process happens rarely enough that you can guarantee only one photon in the apparatus at any given time.
If you use a CCD camera as your target you can easily see how many photons are arriving at a given time.
Then you just put filters in front of your source until you see only one photon arriving at the camera. You’ll have an image with a single dot in what looks like a random position.
Collect a few hundred frames of a single photon arriving and then superimpose them and you’ll end up with the familiar diffraction pattern.
I’ve done this in college. Just measure the intensity of the light in Watts and lower it until it’s produced by single photons.
e.g. photons of frequency f have energy e. Path length is x, time of flight is x/c. If there are only single photons, in 1 second there will be 1/(x/c) = c/x photons. So the energy will be ec/x. If your measured energy is below this, the photons must be coming one at a time.
If one single photon, period, is what you’re looking for, you’d have to resort to some of the (maybe not yet possible) tricks described upthread.
But if all you want to do is be sure only one photon is in the apparatus at a time, the threshold isn’t that hard to reach. I remember a thread from the last couple years that discused the number of photons flying around a room at a given moment. The conclusion drawn was that for a bedroom-sized space illuminated by a single low-wattage light bulb, there may only be one photon in the air at any instant. The reason is they move so incredibly fast that by the time the next one is created, the last one has already hit a wall…even though they are produced very close together temporally. Even for very bright lights in large spaces, the number of photons in the air is more like hundreds than the millions or billions I would have intuitively guessed. It’s easy to underestimate just how fast those suckers move.
The upshot is it may not require much attenuation to be statistically “sure” that only one photon is in your experiment at a time. Photons from everyday light sources are more widely spaced than you think.
EDIT: (correction) Here’s the thread I referenced: How do they generate one (only) photon? - Factual Questions - Straight Dope Message Board I wasn’t remembering the truth, I was remembering another poster’s calculation error. It stuck in my mind because it was a more interesting bit of trivia…but fiction is often better than reality.
But that thread still contains some good discussion about this very topic. The attenuation necessary is a lot more than I indicated above, but not hard to achieve.
This is exactly the question I was looking to get answered. Reading about the double slit experiment, they always state so matter-of-factly that a “single photon” is fired at the slit and detected on the other side. I don’t see how it’s possible to make any photon source that would fire a single photon at a time, let alone a detector that could detect a single photon at a time. Both source and detector seem highly implausible. From the answers thus far (only focused on the source), it seems like they’re based on statistics, rather than actually segregating or producing a single photon.
I don’t think a CCD necessarily reacts to ONE photon. A very sensitive one will react to less photons than a less sensitive one (we get into measurements of lux, resolution, and those superpowerful CCDs used on satellites and in some telescopes, etc). But that doesn’t mean it’s one photon that is registering on the CCD.
The double-slit experiment is mindboggling and amazing, but every description of it that I’ve ever read just takes for granted and glosses over the supposed fact that we can both generate and detect a single photon, one at a time. But can we?
Upthread it’s been described how you can ensure that there is only a single photon in the apparatus at a time. It is also possible to detect single photons using a photomultipliertube.
In school, I did data analysis for an experiment that shot a pulsed laser straight up and used a photomultiplier to detect the photons coming back between pulses (as a measure of the density of the atmosphere at different levels). Counts from 100km up were routinely in the low single digits. In fact, where the counts were into the tens of thousands, we had to apply a correction factor because that was so bright that the tube was partially blinded.
Don’t underestimate the sensitivity of the available equipment.
Single photon detection is routine. It is possible with a CCD or with a number of other technologies (e.g., photomultiplier tubes and Gieger-mode avalanche photodiodes… but there are many approaches to photon detection).
Single photon generation is also routine but, as has been mentioned, is stochastic. To wit:
Consider a 1 mW green laser. The energy of each photon is 3.7x10[sup]-19[/sup] J, which means there are 2.7x10[sup]15[/sup] photons coming out each second. Now place a very dark filter in the beam, perhaps one with only 0.1% transmission. The beam downstream of this filter has only 2.7x10[sup]12[/sup] photons passing a given point each second. Now place 4 more filters in the beam. This brings the photon rate down to 2.7 per second. If your detector has a frame rate of 1/60[sup]th[/sup] second, then each frame will have, on average, 0.045 photons detected.
But one cannot detect a fraction of a photon. What this 0.045 number means is that most frames will be empty, and roughly 1-in-20 frames will have one photon in them, and on rare occasion a frame will have two or more photons in it. This is a textbook example of a Poisson process, and the probabilities for getting zero, one, two, …, photons in a particular frame are given by the Poisson distribution. For the numerical example above:
Probability that the frame is empty = P(0) = 0.956
Probability that the frame has one photon = P(1) = 0.043
P(2) = 0.001
P(3 or above) = much smaller still
If you have some way to identify and discard the empty (zero-photon frames), you will be left with two types of frames:
98%: single-photon frames (what you want!)
2%: two-or-more-photon frames (bad!)
Notice the optimization you can do here. If a 2% multi-photon contamination doesn’t affect the question you’re trying to address, you’re all set. If you need a lower level of contamination, you can just add more filters to your laser to get the single-photon fraction as close to 100% as you need. The only cost is in how many zero-photon frames you’re forced to collect and discard.
You don’t have to stick with the “60 Hz video frames” sort of story here. You could have a single-shot or continuously integrating detector and have the laser be pulsed. So, maybe the laser emits a single short pulse with 10[sup]9[/sup] photons in it. Simple attenuate that number down to whatever you need using filters, and you can obtain an arbitrarily low probability of getting more than one photon, at the expense of an increasingly high probability of getting zero photons.
Black-and-white visible-light CCDs routinely manage to get quantum efficiencies of upwards of 90%. That means that, for each individual photon that hits it, you’re 90% likely to detect it. So, yeah, you’ll miss a few, but it’s still pretty accurate to say you’re detecting individual photons.
Do any of these detectors measure energy as well? If photons travel in pairs, would the detectors know?
The most common detector types rely on an incoming photon to kick an electron out of a metal or across a semiconductor band gap, sometimes with a high voltage bias present to encourage the bumped electron to start a cascade, thus providing amplification. For these, the output is proportional to the number of photons detected, and depending on the exact design, the resolution on “number of photons” can range from poor to excellent. (In the excellent cases, you can reliably say, for instance, that 5 photons were detected and not 4 or 6.) None of these tell you the energy of the incoming photons, just how many there were (except via the fact that they are only sensitive to a particular range of photon wavelengths, but that’s usually not the point.)
Sometimes the application at hand warrants running with a very high bias voltage, with the side effect that you can only tell whether any photon(s) were detected, but you can’t say how many.
If you do require an energy measurement for a single photon, you can get this with somewhat more cumbersome cryogenic devices. One choice is a transition edge sensor, a calorimeter that uses a cryogenic superconductor operating at its transition threshold. A small energy deposition from a single photon will increase the temperature of the device slightly, and given the extreme dependence of resistance on temperature at the superconducting transition, a measurement of the blip of an increase in resistance can tell you how much energy was deposited.
It is worth nothing that the discussion in this thread has focussed on visible-ish photons. For x-rays and gamma rays, single photon creation and detection (and energy measurement) is also routine, but it starts to look very different.
I mean, wouldn’t photons traveling in pairs (like electrons in superconductivity) be a simpler explanation for this quantum weirdness? I can’t think of any experiments that show photons DON’T travel in pairs right now.
We don’t just do two-slit experiments. You can just as easily do a three-slit, or ten-slit, or hundred-slit interference experiment, so each “photon bundle” would have to have a great many photons in it to account for the results. And all light has the same relationship between wavelength and the size of the energy quantum, which would mean that all photon bundles would have to contain the same number of photons. And the constant that shows up in that relationship shows up everywhere else in quantum mechanics, too, so you’d also have to come up with some way of accounting for the spectrum of the hydrogen atom, and the blackbody spectrum, and so on. And if the photon bundles can split up to go through the multiple slits, what causes them to join back up before hitting the screen, and why don’t we ever observe them split up?
Isn’t 2 photons enough for diffraction regardless of number of slits? That’s the reason I asked how sensitive our detectors are - maybe they never split up and we can’t tell the difference between 3E-19 J every 3E-16 seconds, and 6E-19 J every 6E-16 seconds.
Then again, I think they’ve done diffraction using large ions/molecules which they can be sure pass 1 at a time, so maybe not.